New study compares MCP vs A2A for AI agent coordination

A new academic paper from researchers at the University of York presents an implementation-grounded comparison of two emerging protocols for inter-agent coordination in LLM-based systems: the Model Context Protocol (MCP) and Agent2Agent (A2A). The study, submitted to arXiv in July 2026, evaluates both protocols against a set of requirements including agent discoverability, multi-part messaging, multi-turn conversations, asynchronous communication, observability, interoperability, and access control. The researchers implemented the same software engineering task using both approaches and analyzed the trade-offs.

The results show that MCP can support inter-agent coordination in constrained LLM-based systems through a comparatively lightweight implementation model with lower coordination complexity. However, coordination concerns such as conversational state management and task lifecycle handling must be implemented explicitly at the application layer when using MCP. In contrast, A2A provides richer native support for stateful, multi-turn coordination through protocol-level abstractions for tasks and lifecycle management, but this comes with substantially greater implementation and coordination complexity. The authors emphasize these findings are design observations from an empirical experience report, not general claims of protocol superiority, highlighting how protocol abstractions shape distribution of coordination responsibilities in contemporary agentic systems.